Abstract

Pollution sources are determined by source apportionment, and are used as human-influenced factors to calculate the pollution risk of regional groundwater. In this case study, 266 samples of shallow groundwater were collected from the Lower Liaohe River Plain in northeastern China. Hydrochemical indexes of NH4+, Ca2+, Cl, F, HCO3, Pb, Mg2+, NO3, NO2, K+, Na+, SO42−, total Cr, chemical oxygen demand, total dissolved solids, and total hardness of each sample were determined. Factor analysis method was used to identify potential pollution sources. Five common factors (agricultural activities, groundwater extraction, industrial activities, river pollution, and sewage discharges) represented 77.10% of the total variance. The results showed that the relatively high pollution risk mainly occurred in the west, with 19.91% of the area. The southeastern area showed a low risk of pollution, covering only 1.07% of the study area. Clearly, human activities have significantly changed the distribution of regional groundwater pollution risk in the study area.

INTRODUCTION

Groundwater is one of the most important sources of water supply worldwide (Zhao & Pei 2012), and it is under intense anthropogenic influences and rising pollution threat. Human activities, such as agriculture and industry, and urbanization, have caused groundwater to be degraded irreversibly (Kazakis & Voudouris 2015). Groundwater risk assessment studies are becoming increasingly popular in the search for effective ways of evaluating the sensitivity of groundwater pollution. Groundwater pollution risk reflects the sensitivity of the regional groundwater to human activities and natural occurrences, which could be defined as the possibility that contaminants enter groundwater from surface water (Babiker et al. 2005). Groundwater risk is closely related to two factors: intrinsic vulnerability and human activity. Intrinsic vulnerability can be defined as a contaminant introduced into the surface and being diffused in groundwater, and human activity is the vulnerability of groundwater to a particular or group of contaminants, taking into account the properties of the contaminant and the relationships between contaminants (Gogu & Dassarguesm 2000). The DRASTIC model is the most widely used overlay index method to assess the groundwater vulnerability (Fritch et al. 2000). Al-Adamat et al. (2003) calculated the vulnerability with the DRASTIC index, using GIS to identify potentially contaminated areas. And particularly noteworthy is that Ouedraogo et al. (2015), who estimated the vulnerability and pollution risk of groundwater at the pan-African scale, which was the technical improvement of the large scale study on the vulnerability and risk assessment of groundwater.

Pollution risk of regional groundwater is regarded as being more meaningful than the intrinsic vulnerability, because some factors have affected the intrinsic vulnerability and are continuing to be changed due to human activities (Huan et al. 2012). Even though human activities are commonly used as indicators of groundwater vulnerability (Lytton et al. 2003; Huan et al. 2012), nitrogen as ammonium has been found to be representative of groundwater quality (Iqbal et al. 2015). With the aim of improving the accuracy of risk assessment of groundwater pollution, many different parameters and weightings have been adopted and conducted under various conditions (Panagopoulos et al. 2006; Bai et al. 2016), but the risk of groundwater still cannot be fully evaluated from water quality. A more representative factor of water quality should be selected to evaluate groundwater risk effectively. Multivariate statistical analysis was used to identify factors that could better represent water quality than is currently possible.

Multivariate analysis includes regression analysis, cluster analysis, principal component analysis, factor analysis (FA) and so on. It has been widely used for groundwater quality characteristics and the evolution of hydro-geochemical distributions. Moya et al. (2015) used cluster analysis, FA, and principal component analysis to identify pollution sources and the processes involved in the evolution of major aquifers in the Galilee and Eromanga basins. Furthermore, multivariate statistical analyses have been widely used in areas including the function of rivers, groundwater and marine waters to identify the pollution sources, and have provided important reference for the management of groundwater resources (Alberto et al. 2001; Papatheodorou et al. 2006). Hynds et al. (2014) gathered 211 wells sampled within the Republic of Ireland, and combining with the analysis of geological conditions, found that the incidence of bacteria in groundwater is mainly affected by septic tanks and agricultural activities. Machiwal & Jha (2015) combined ArcGIS analysis and principal component analysis to determine the contribution of natural/geological causes and anthropogenic sources to groundwater pollution in hilly areas of Udaipur, India.

In this study, groundwater samples from shallow groundwater in the Lower Liaohe River Plain were used to identify the main factors that influenced the water quality by multivariate statistical analysis. These factors were used as pollution risk sources, combined with the DRASTIC model, the regional pollution risk was calculated.

MATERIALS AND METHODS

Study area

The study area was located in the Lower Liaohe River Plain in central Liaoning Province of northeastern China, covered about 23,470 km2 and included nine cities such as Anshan, Fushun, Shenyang and so on, as shown in Figure 1. The terrain gently increases with elevation from north to south and to both sides until the transition areas of Liaodong and Liaoxi are reached. The Daling River, Hunhe River, Liaohe River, Raoyang River, Shuangtaizi River, and Xiaoling River converge in the study area and then drain into the Bohai Sea. The mean annual rainfall is 623.2 mm, and the mean annual evaporation is 1,669.6 mm. The Lower Liaohe River Plain receives recharge from surface runoff, and subsurface runoff occurs in the eastern, western, and northern parts of the plain, which are rich in groundwater. The main discharge routes are horizontal and vertical, and the discharge routes have been affected strongly by Quaternary transgression. Groundwater in the Quaternary strata south of Panshan Mountain is saline, and the hydrogeological zonation conditions have led to salinity of the shallow groundwater and the hydro-chemistry to be appreciably different in different zones. Particular conditions in certain areas cause bicarbonate sulfate and sulfate bicarbonate water types to appear spontaneously.

Figure 1

Hydrogeological conditions and sampling points in the study area.

Figure 1

Hydrogeological conditions and sampling points in the study area.

Analytical methods

All of the samples were collected from the shallow groundwater aquifer in the Lower Liaohe River Plain. A total of 266 samples were collected from individual wells or irrigation wells; the locations of samples are shown in Figure 1. Groundwater was sampled from each well either by hand-pressure or by using a submersible pump. The samples were collected between September 2009 and October 2010. In this study, data pre-treatment methods, such as the elimination of non-informative variables, the treatment of missing data values, and the detection and treatment of outliers were performed before FA. The temperature, pH and electrical conductivity (EC) were measured in-situ, which were measured with calibrated EC and pH meters. Each sample was filtered by a 0.45 μm filter membrane and put in a 125 mL container, and then placed in a high-density polyethylene bottle with 2.5 mL nitric acid (1:1) and stored at 4 °C until analysis. Each sample was analyzed less than one week after being collected. , Ca2+, Cl, F, , Pb2+, Mg2+, , , K+, Na+, , total Cr concentrations, chemical oxygen demand (COD), total dissolved solid (TDS), and total hardness (TH) were analyzed in the laboratory. The analytical method is presented in Table 1. The relative error was less than ±5% for all analyzed indexes.

Table 1

Analytical methods for groundwater quality parameters

ParameterStandard methods and techniquesParameterStandard methods and techniques
COD Potassium dichromate method ,  Ultraviolet spectrophotometer 
TDS Gravimetric method by drying at 105°C  Acid base titration 
TH Ethylene Diamine Tetraacetic Acid titration , , F, Cl Ion chromatographic 
Pb, Total Cr Inductively coupled plasma-mass spectrometry K+, Na+, Ca2+, Mg2+ Inductively coupled plasma-mass spectrometry 
ParameterStandard methods and techniquesParameterStandard methods and techniques
COD Potassium dichromate method ,  Ultraviolet spectrophotometer 
TDS Gravimetric method by drying at 105°C  Acid base titration 
TH Ethylene Diamine Tetraacetic Acid titration , , F, Cl Ion chromatographic 
Pb, Total Cr Inductively coupled plasma-mass spectrometry K+, Na+, Ca2+, Mg2+ Inductively coupled plasma-mass spectrometry 

FA was used to determine the contributions of different natural and anthropogenic factors. The data were processed and analyzed using IBM SPSS Statistics Version 20.0 software (IBM, Armonk, NY, USA). Two problems were found and processed, the first being that the effects of the sampling methods and the precision of the analytical methods caused most of the test results to be below the detection limits, meaning that principal component analysis could not be performed on the data, and the second being that agricultural activities occur frequently in the Lower Liaohe River Plain, and the Liaohe Delta is affected by long-term seawater intrusion. The above 16 indexes were selected, and the results are shown in Table 2 and Figure 2.

Table 2

Statistical summary of the water quality parameter results

Water quality parameter (mg/L)MinimumMaximumMeanStandard deviationVariance
Ca2+ 8.89 995.40 118.64 104.67 10,956.11 
Mg2+ 3.00 575.00 52.03 75.73 5,735.32 
K+ 0.08 369.70 10.93 32.20 1,036.71 
Na+ 5.55 1,125.00 134.47 202.49 41,002.23 
Cl 0.02 3,712.14 205.39 441.67 195,070.24 
 2.00 4,966.11 161.88 447.53 200,280.14 
 49.40 1,924.30 386.97 277.29 76,887.29 
 0.20 276.00 19.89 40.83 1,666.80 
 0.01 3.62 0.09 0.37 0.14 
 0.02 233.50 3.06 18.56 344.44 
COD 0.50 80.55 4.94 7.79 60.70 
TH (CaCO341.22 4,656.85 516.89 542.79 294,626.06 
TDS 196.50 8,520.72 1,161.49 1,157.56 1,339,945.72 
F <DL 7.50 0.51 0.73 0.53 
Pb <DL 0.10 0.0057 0.01 0.00 
Total Cr <DL 0.10 0.0098 0.01 0.00 
Water quality parameter (mg/L)MinimumMaximumMeanStandard deviationVariance
Ca2+ 8.89 995.40 118.64 104.67 10,956.11 
Mg2+ 3.00 575.00 52.03 75.73 5,735.32 
K+ 0.08 369.70 10.93 32.20 1,036.71 
Na+ 5.55 1,125.00 134.47 202.49 41,002.23 
Cl 0.02 3,712.14 205.39 441.67 195,070.24 
 2.00 4,966.11 161.88 447.53 200,280.14 
 49.40 1,924.30 386.97 277.29 76,887.29 
 0.20 276.00 19.89 40.83 1,666.80 
 0.01 3.62 0.09 0.37 0.14 
 0.02 233.50 3.06 18.56 344.44 
COD 0.50 80.55 4.94 7.79 60.70 
TH (CaCO341.22 4,656.85 516.89 542.79 294,626.06 
TDS 196.50 8,520.72 1,161.49 1,157.56 1,339,945.72 
F <DL 7.50 0.51 0.73 0.53 
Pb <DL 0.10 0.0057 0.01 0.00 
Total Cr <DL 0.10 0.0098 0.01 0.00 

DL, detection limit.

Figure 2

Distribution of the concentration of each index in the study area.

Figure 2

Distribution of the concentration of each index in the study area.

Groundwater geochemical processes and heterogeneity caused by human activities was analyzed with an inverse distance weighted interpolation (Figure 2).

Factor analysis

FA is one of the most important multivariate statistical methods, and it is an internal analysis using a lower number of unobserved variables called factors to explain more complex relationships in the observed variables. The basic idea of FA is to identify correlations between variables and then to group the variables together based on the identified correlations. Each group of variables represents a basic structure that could potentially be expressed using common factors. The FA could be expressed as:  
formula

Matrix, X = AF + E,

Where,

  • Xj are the standardized scores for the variables,

  • Fi (i = 1, 2, … , m) are the common factors,

  • m is the number of common factors for all of the variables,

  • ɛj (j = 1, 2, … , p) are the specific factors,

  • aji are the factor loadings.

FA requires that three conditions should be met: (1) the common factors and specific factors are not related; (2) each common factor is not related to another, and the variance of each common factor is 1; and (3) each special factor is not related to another, and the variances of the specific factors are not required to be equal. When the pollution source groups are more than 10, FA almost cannot provide good recognition results (Gordon 1988).

Groundwater vulnerability model

The DRASTIC model is composed of seven hydrogeological factors, and was calculated by linearly combining all of the indexes according to the following equation:  
formula
where DI is the vulnerability index based on the DRASTIC model, D is the depth to the groundwater, R is the recharge of regional groundwater, A is the aquifer type, S is the soil media type, T is the topography, I is the impact of the vadose zone, C is the hydraulic conductivity, and λindex is the weighting for the factor indicated by the corresponding subscript. Higher DI indicates that the groundwater is more vulnerable, and the groundwater vulnerability mapping could be graded into several classifications indicating comparable possibilities of groundwater pollution.

RESULTS AND DISCUSSION

Description of the correlations

The correlation matrix for the 16 groundwater indexes from the 266 groundwater samples is shown in Table 3. Ca2+, Cl, Mg2+, TDS and TH were correlated strongly, indicating that they have all come from the same sources. The concentration of K+ was closely related to , which indicated that they came from the same sources. Meanwhile, there were a good correlation between , Pb, COD and Cr, which showed that they came from the same sources. The correlation between and other parameters is poor or negative, indicating that it originated from a different source from the other parameters.

Table 3

Correlation matrix for the water quality parameters

 Ca2+Mg2+K+Na+ClCODTHTDSFPbTotal Cr
Ca2+                
Mg2+ 0.766**               
K+ 0.359** 0.682**              
Na+ 0.470** 0.732** 0.587**             
Cl 0.708** 0.636** 0.237** 0.732**            
 0.402** 0.721** 0.817** 0.511** 0.086           
 0.307** 0.505** 0.339** 0.552** 0.286** 0.241**          
 0.214** 0.062 −0.034 −0.082 −0.042 0.06 −0.184**         
 0.125* 0.289** 0.441** 0.322** 0.061 0.41** 0.185** −0.006        
 0.028 0.022 0.103* −0.004 0.004 −0.014 0.359** −0.055 −0.033       
COD 0.298** 0.477** 0.62** 0.429** 0.21** 0.496** 0.452** −0.045 0.429** 0.57**      
TH 0.927** 0.95** 0.57** 0.653** 0.713** 0.613** 0.441** 0.137* 0.23** 0.029 0.422**     
TDS 0.758** 0.94** 0.697** 0.888** 0.722** 0.68** 0.602** 0.043 0.33** 0.087 0.533** 0.912**    
F −0.053 0.004 −0.027 0.031 −0.074 0.047 0.089 −0.056 0.009 0.096 −0.027 −0.024 0.015   
Pb 0.086 0.001 −0.046 −0.091 −0.025 −0.032 −0.02 0.063 0.023 0.093 0.184** 0.014 0.042 −0.016  
Total Cr 0.034 −0.026 −0.055 −0.003103 −0.046 −0.029 −0.036 −0.053 0.021 0.1* 0.132* 0.029 0.005 −0.046 0.383** 
 Ca2+Mg2+K+Na+ClCODTHTDSFPbTotal Cr
Ca2+                
Mg2+ 0.766**               
K+ 0.359** 0.682**              
Na+ 0.470** 0.732** 0.587**             
Cl 0.708** 0.636** 0.237** 0.732**            
 0.402** 0.721** 0.817** 0.511** 0.086           
 0.307** 0.505** 0.339** 0.552** 0.286** 0.241**          
 0.214** 0.062 −0.034 −0.082 −0.042 0.06 −0.184**         
 0.125* 0.289** 0.441** 0.322** 0.061 0.41** 0.185** −0.006        
 0.028 0.022 0.103* −0.004 0.004 −0.014 0.359** −0.055 −0.033       
COD 0.298** 0.477** 0.62** 0.429** 0.21** 0.496** 0.452** −0.045 0.429** 0.57**      
TH 0.927** 0.95** 0.57** 0.653** 0.713** 0.613** 0.441** 0.137* 0.23** 0.029 0.422**     
TDS 0.758** 0.94** 0.697** 0.888** 0.722** 0.68** 0.602** 0.043 0.33** 0.087 0.533** 0.912**    
F −0.053 0.004 −0.027 0.031 −0.074 0.047 0.089 −0.056 0.009 0.096 −0.027 −0.024 0.015   
Pb 0.086 0.001 −0.046 −0.091 −0.025 −0.032 −0.02 0.063 0.023 0.093 0.184** 0.014 0.042 −0.016  
Total Cr 0.034 −0.026 −0.055 −0.003103 −0.046 −0.029 −0.036 −0.053 0.021 0.1* 0.132* 0.029 0.005 −0.046 0.383** 

The TH was determined from the CaCO3 concentration.

*Indicates the correlation was significant at the 95% confidence interval.

**Indicates the correlation was significant at the 99% confidence interval.

Identification of pollution sources

The data were standardized to give a mean of 0 and a variance of 1 before FA was performed. The Kaiser–Meyer–Olkin (KMO) and the Bartlett Test of Sphericity were applied to verify the suitability of the data for FA, and the calculation results are shown in Table 4. According to the results, the KMO value was 0.610 and the Bartlett test was significant. All those results indicated that the data were suitable for FA. A total of five common factors were extracted to determine the number of principal components with the principle of eigenvalues >1, as shown in Table 5.

Table 4

KMO and Bartlett test results

Sampling sufficient degree of KMO measure0.610
Bartlett's sphericity test Degrees of freedom (d.f.) 120 
Significance (Sig.) 0.000 
Sampling sufficient degree of KMO measure0.610
Bartlett's sphericity test Degrees of freedom (d.f.) 120 
Significance (Sig.) 0.000 
Table 5

Eigenvalues of the common factors and the proportions of the variance of the data explained by the factors

FactorInitial eigenvalue
Extracting sum of squares
Rotating sum of squares
TotalVariance %Cumulative %TotalVariance %Cumulative %TotalVariance %Cumulative %
6.471 40.441 40.441 6.471 40.441 40.441 4.921 30.756 30.756 
1.809 11.308 51.749 1.809 11.308 51.749 2.903 18.143 48.899 
1.557 9.729 61.478 1.557 9.729 61.478 1.736 10.852 59.751 
1.386 8.662 70.140 1.386 8.662 70.140 1.524 9.527 69.278 
1.114 6.960 77.100 1.114 6.960 77.100 1.251 7.822 77.100 
0.919 5.743 82.843       
0.713 4.459 87.302       
0.595 3.719 91.021       
0.538 3.364 94.385       
10 0.450 2.811 97.196       
11 0.185 1.158 98.355       
12 0.163 1.017 99.372       
13 0.093 0.578 99.950       
14 0.006 0.038 99.988       
15 0.001 0.007 99.994       
16 0.001 0.006 100.000       
FactorInitial eigenvalue
Extracting sum of squares
Rotating sum of squares
TotalVariance %Cumulative %TotalVariance %Cumulative %TotalVariance %Cumulative %
6.471 40.441 40.441 6.471 40.441 40.441 4.921 30.756 30.756 
1.809 11.308 51.749 1.809 11.308 51.749 2.903 18.143 48.899 
1.557 9.729 61.478 1.557 9.729 61.478 1.736 10.852 59.751 
1.386 8.662 70.140 1.386 8.662 70.140 1.524 9.527 69.278 
1.114 6.960 77.100 1.114 6.960 77.100 1.251 7.822 77.100 
0.919 5.743 82.843       
0.713 4.459 87.302       
0.595 3.719 91.021       
0.538 3.364 94.385       
10 0.450 2.811 97.196       
11 0.185 1.158 98.355       
12 0.163 1.017 99.372       
13 0.093 0.578 99.950       
14 0.006 0.038 99.988       
15 0.001 0.007 99.994       
16 0.001 0.006 100.000       

Table 5 explains 77.10% of the total variance, which indicates that it could well represent the original quality characteristics of the data. The five common factors represent 30.76%, 18.14%, 10.85%, 9.53%, and 7.82% of the total variance, respectively. The rotation ingredient matrix (also known as the factor loading matrix) is shown in Table 6.

Table 6

Rotation ingredient matrix

Water quality parameterFactor
F1F2F3F4F5
Ca2+ 0.870 0.092 0.049 0.095 0.294 
Mg2+ 0.835 0.467 0.077 0.005 0.036 
K+ 0.358 0.810 0.171 −0.077 −0.022 
Na+ 0.750 0.368 0.066 −0.133 −0.302 
Cl 0.902 −0.138 0.008 −0.069 −0.097 
 0.334 0.851 0.009 0.001 0.087 
 0.463 0.158 0.517 −0.038 −0.386 
 0.062 0.020 −0.051 0.021 0.842 
 0.028 0.720 0.017 0.041 −0.068 
 −0.035 −0.061 0.941 0.107 0.001 
COD 0.232 0.539 0.708 −0.014 0.008 
TH (CaCO30.905 0.315 0.071 0.050 0.161 
TDS 0.871 0.445 0.155 −0.031 −0.059 
F −0.028 0.065 −0.043 0.513 −0.381 
Pb 0.024 −0.033 0.048 0.793 0.136 
Total Cr −0.026 −0.019 0.055 0.759 0.006 
Water quality parameterFactor
F1F2F3F4F5
Ca2+ 0.870 0.092 0.049 0.095 0.294 
Mg2+ 0.835 0.467 0.077 0.005 0.036 
K+ 0.358 0.810 0.171 −0.077 −0.022 
Na+ 0.750 0.368 0.066 −0.133 −0.302 
Cl 0.902 −0.138 0.008 −0.069 −0.097 
 0.334 0.851 0.009 0.001 0.087 
 0.463 0.158 0.517 −0.038 −0.386 
 0.062 0.020 −0.051 0.021 0.842 
 0.028 0.720 0.017 0.041 −0.068 
 −0.035 −0.061 0.941 0.107 0.001 
COD 0.232 0.539 0.708 −0.014 0.008 
TH (CaCO30.905 0.315 0.071 0.050 0.161 
TDS 0.871 0.445 0.155 −0.031 −0.059 
F −0.028 0.065 −0.043 0.513 −0.381 
Pb 0.024 −0.033 0.048 0.793 0.136 
Total Cr −0.026 −0.019 0.055 0.759 0.006 

The first common factor (F1) primarily consisted of Ca2+, Cl, Mg2+, Na+, TDS, and TH. At least three transgression events have occurred in the Lower Liaohe River Plain since the late Pleistocene. Each transgression event had an important effect on groundwater chemical evolution of the area, and there remain many kinds of coastal geomorphological features (Liu et al. 2008; Rossetti et al. 2013). As shown in Figure 2(a), the above indexes were relatively high in groundwater from the southern coastal plain. Erban & Gorelick (2016) showed that the large-scale exploitation and overexploitation of groundwater resources could cause variations in the regional groundwater flow field. Qin et al. (2013) showed that Ca2+ and Cl, TDS, and TH can change significantly under different pumping conditions. Tomaszkiewicz et al. (2014) studied the evolution of groundwater with respect to mining conditions around the Bohai Rim, and found that the exploitation and overexploitation of groundwater has exacerbated the intrusion of seawater into coastal aquifers. When the balance between saltwater and freshwater is broken, the concentrations of Ca2+, Cl, Na+ and TH in groundwater would increase.

F2 mainly consisted of , K+, and . As is shown in Figure 3(b), high F2 values were primarily within the Shuangtaizi River basins, where rivers and groundwater have a strong regional exchange. Moreover, the groundwater depths of the sampling points showed that the groundwater was relatively shallow and would easily be mixed with surface water in this area, and intensive agricultural activities occurred. The excessive application of potassium sulfate can cause high K+ and in the surface environment easily entering into rivers from flooding and paddy irrigation water overflows. Denitrifying bacteria in river sediments could increase the nitrite content of the water. Overall, the distribution of surface water, river–groundwater exchange, and denitrification would have contributed strongly to F2.

Figure 3

Kriging space interpolation for each common factor.

Figure 3

Kriging space interpolation for each common factor.

F3 mainly consisted of and COD. It has been shown that sewage discharges increase the concentration of ammonium and COD in rivers and groundwater. As shown in Figure 3(c), higher F3 values were mainly found in the northern Piedmont region, around Shenyang City, and on the southern coastal plain. The high F3 value in the Shenyang City area was similar to Hunhe River, which received urban sewage. Downstream areas of the Lower Liao River and Shuangtaizi River also had high F3 values. F3 should be attributed to the effects of sewage discharges.

F4 mainly consisted of Pb and Cr. As shown in Figure 3(d), high F4 were mainly in the northern Lower Liaohe River Plain. The eastern and northwestern mountain area was the major recharge area in the study area; after mine exploitation and rock weathering, Cr and Pb released from the rock would move downstream through long-term eluviation and transportation. Therefore, Cr and Pb in groundwater would dissolve from minerals containing the two metals. However, industrial activities, such as mining, and sewage effluents have altered the original geological environment, accelerating the leaching of Cr and Pb, so F4 was attributed to industrial activities (mainly in chemical plants).

F5 mainly consisted of . It can be seen from Figure 3(e) that high F5 values covered a large area of the Lower Liaohe River Plain, particularly in the western Piedmont and northern plain, where frequent agricultural activities occurred. Long-term agricultural activities cause nitrogen as nitrate to infiltrate to groundwater through eluviation. Li et al. (2015) confirmed that agricultural activities affect the regional groundwater pollution risk. F5 was therefore attributed to agricultural activities.

Spatial distribution of pollution

The kriging interpolation method was applied to obtain the distributions of the factors using ArcGIS 9.3. A higher factor value would have been associated with more severe regional pollution. As shown in Figure 3, different pollution sources caused different degrees of contamination in different areas of the Lower Liaohe River Plain.

Pollution risk assessment of groundwater

The pollution risk of groundwater has two components, the intrinsic vulnerability and the anthropogenic impacts on the groundwater.

(1) Intrinsic vulnerability of groundwater

The DRASTIC model was used to calculate the groundwater intrinsic vulnerability, and the results are shown in Figure 4.

(2) Human impacts

The five factors were considered to be human impacts on pollution, and were groundwater extraction, pollutant concentrations from rivers, sewage discharges, chemical enterprises, and agricultural activities respectively (Figure 5).

Figure 4

Intrinsic vulnerability of groundwater calculated using the DRASTIC model.

Figure 4

Intrinsic vulnerability of groundwater calculated using the DRASTIC model.

Each factor had a different impact on the groundwater vulnerability with a different weight. Little scientific evidence that could be used to determine an appropriate weighting for each factor was available, so it could be obtained using the results of FA analysis. The five factors that are shown in Table 5 explained 77.10% of the total variance, indicating that the results could be used to represent the characteristics of the raw water quality. The weighting for each factor was determined according to the contributed percentage to the total variance (shown in Table 5). The normalization process gave the weightings for the five factors shown in Table 7. The sum of the weightings was 1.

Figure 5

Distribution of the normalization factors.

Figure 5

Distribution of the normalization factors.

Table 7

Factor weightings

FactorWeighting
Groundwater extraction 0.40 
Pollutants from river water 0.24 
Sewage discharge 0.14 
Chemical enterprises 0.12 
Agricultural activities 0.10 
FactorWeighting
Groundwater extraction 0.40 
Pollutants from river water 0.24 
Sewage discharge 0.14 
Chemical enterprises 0.12 
Agricultural activities 0.10 
(3) Pollution risk of regional groundwater

The calculated intrinsic vulnerability was combined with human impacts to determine the regional groundwater pollution risk. The intrinsic vulnerability determined using the DRASTIC model was an objective result, and human factors were actual values (determined through calculations). The following three steps were taken to obtain the results, using the Grid computing system in ArcGIS 9.3: (1) the intrinsic vulnerability results were normalized; (2) the five human factors were normalized and then multiplied by their respective weightings, then the results were combined; and (3) the final results of the two previous steps were combined. The calculated results of pollution risk are shown in Figure 6.

Figure 6

Pollution risk of regional groundwater in the study area.

Figure 6

Pollution risk of regional groundwater in the study area.

The results shown in Figure 6 and Table 8 could draw the following conclusions.

Table 8

Differences between the pollution risk and intrinsic vulnerability distributions

 Proportion of the study area (%)
Area (km2)
Intrinsic vulnerabilityPollution riskVariation widthIntrinsic vulnerabilityPollution riskVariation
Low 2.43 1.07 −1.36 569.63 250.37 −319.26 
Relatively low 22.31 14.67 −7.64 5,236.07 3,442.16 −1,793.91 
Moderate 54.93 45.09 −9.84 12,892.02 10,581.73 −2,310.29 
Relatively high 19.38 19.91 +0.53 4,547.83 4,673.91 +126.08 
High 0.96 19.27 +18.31 224.44 4,521.84 +4,297.4 
 Proportion of the study area (%)
Area (km2)
Intrinsic vulnerabilityPollution riskVariation widthIntrinsic vulnerabilityPollution riskVariation
Low 2.43 1.07 −1.36 569.63 250.37 −319.26 
Relatively low 22.31 14.67 −7.64 5,236.07 3,442.16 −1,793.91 
Moderate 54.93 45.09 −9.84 12,892.02 10,581.73 −2,310.29 
Relatively high 19.38 19.91 +0.53 4,547.83 4,673.91 +126.08 
High 0.96 19.27 +18.31 224.44 4,521.84 +4,297.4 

The pollution risk of groundwater in the Lower Liaohe River Plain had not presented a particularly obvious regularity, and the trends were roughly the same as those in the intrinsic vulnerability. It can be seen from Figure 6 that high pollution risk mainly occurred in the southwest, southeast, and northeast of the study area, and the total area of high pollution risk was smaller than the other areas. Relatively high pollution risk mainly occurred in the western part of the study area, and accounts for 19.91% of the total area. Moderate pollution risk was widely distributed over the study area, and accounted for the largest area in this region. Relatively low pollution risk mainly occurred at the edge of the study area, especially surrounding the Daliaohe and Shuangtaizihe rivers. Low pollution risk mainly occurred in the southeastern part of the study area, and only accounted for 1.07% of the total area. So, human activities had significantly changed the distribution of pollution risk of regional groundwater.

CONCLUSIONS

Based on the apportionment of pollution sources for groundwater in the Lower Liaohe river plain, five common factors with eigenvalues >1, explaining 77.10% of the total variance in the data, were extracted. The common factors of F1, F2, F3, F4 and F5 represented groundwater extraction, river pollution, sewage discharge, chemical enterprises, and agricultural activities, respectively. The pollution risk assessment of regional groundwater showed that high risk mainly occurred in the southwestern, southeastern, and northeastern parts of the study area; relatively high risk mainly occurred in the western part of the study area; moderate risk was found throughout the study area; relatively low risk was mainly at the edges of the study area; meanwhile, the low risk was mainly in the southeastern part of the study area. Combining the distributions of intrinsic vulnerability with human impacts showed that human activities had significantly changed the distribution of risk in the Lower Liaohe River Plain. And when the described water quality is available, FA analysis could be used as an effective method for evaluating the pollution risk of groundwater.

ACKNOWLEDGEMENTS

This work was supported by the National Natural Science Foundation of China (No. 41402211 and No. 41372233), and the Major Science and Technology Program for Water Pollution Control and Treatment (2014ZX07201-010).

REFERENCES

REFERENCES
Alberto
,
W. D.
,
Pilar
,
D. M.
,
Valeria
,
A. M.
,
Fabiana
,
P. S.
,
Cecilia
,
H. A.
&
Ángeles
,
B. M.
2001
Pattern recognition techniques for the evaluation of spatial and temporal variations in water quality
.
A case study: Suquía River Basin (Córdoba–Argentina)
.
Water Research
35
(
12
),
2881
2894
.
Babiker
,
I. S.
,
Mohamed
,
M. A. A.
,
Hiyama
,
T.
&
Kato
,
K.
2005
A GIS-based DRASTIC model for assessing aquifer vulnerability in Kakamigahara Heights, Gifu Prefecture, central Japan
.
Science of the Total Environment
345
(
1–3
),
127
140
.
Bai
,
L.
,
Wang
,
Y.
,
Zhou
,
Y.
,
Liu
,
L.
&
Yan
,
Z.
2016
Research on the process-based risk evaluation method of groundwater pollution for contaminated site
.
Water Science and Technology: Water Supply
16
(
1
),
150
162
.
Fritch
,
T. G.
,
McKnight
,
C. L.
,
Yelderman
,
J. C.
&
Arnold
,
J. G.
2000
An aquifer vulnerability assessment of the Paluxy aquifer, central Texas, USA, using GIS and a modified DRASTIC approach
.
Environmental Management
25
(
3
),
337
345
.
Gordon
,
G. E.
1988
Receptor models
.
Environmental Science & Technology
22
(
10
),
1132
1142
.
Hynds
,
P.
,
Misstear
,
B. D.
,
Gill
,
L. W.
&
Murphy
,
H. M.
2014
Groundwater source contamination mechanisms: physicochemical profile clustering, risk factor analysis and multivariate modelling
.
Journal of Contaminant Hydrology
159
,
47
56
.
Iqbal
,
J.
,
Gorai
,
A. K.
,
Katpatal
,
Y. B.
&
Pathak
,
G.
2015
Development of GIS-based fuzzy pattern recognition model (modified DRASTIC model) for groundwater vulnerability to pollution assessment
.
International Journal of Environmental Science and Technology
12
(
10
),
3161
3174
.
Liu
,
J.
,
Li
,
A.
,
Chen
,
M.
,
Xiao
,
S.
&
Wan
,
S.
2008
Sedimentary changes during the Holocene in the Bohai Sea and its paleoenvironmental implication
.
Continental Shelf Research
28
(
10–11
),
1333
1339
.
Lytton
,
L.
,
Howe
,
S.
,
Sage
,
R.
&
Greenaway
,
P.
2003
Groundwater abstraction pollution risk assessment
.
Water Science and Technology
47
(
9
),
1
7
.
Ouedraogo
,
I.
,
Defourny
,
P.
&
Vanclooster
,
M.
2015
Mapping the groundwater vulnerability for pollution at the pan African scale
.
Science of the Total Environment
544
,
939
953
.
Panagopoulos
,
G. P.
,
Antonakos
,
A. K.
&
Lambrakis
,
N. J.
2006
Optimization of the DRASTIC method for groundwater vulnerability assessment via the use of simple statistical methods and GIS
.
Hydrogeology Journal
14
(
6
),
894
911
.
Papatheodorou
,
G.
,
Demopoulou
,
G.
&
Lambrakis
,
N.
2006
A long-term study of temporal hydrochemical data in a shallow lake using multivariate statistical techniques
.
Ecological Modeling
193
(
3–4
),
759
776
.
Rossetti
,
D. F.
,
Bezerra
,
F. H. R.
&
Dominguez
,
J. M. L.
2013
Late Oligocene–Miocene transgressions along the equatorial and eastern margins of Brazil
.
Earth-science Reviews
123
,
87
112
.
Tomaszkiewicz
,
M.
,
Najm
,
M. A.
&
El-Fadel
,
M.
2014
Development of a groundwater quality index for seawater intrusion in coastal aquifers
.
Environmental Modelling & Software
57
,
13
26
.
Zhao
,
Y.
&
Pei
,
Y.
2012
Risk evaluation of groundwater pollution by pesticides in China: a short review
.
Procedia Environmental Sciences
13
,
1739
1747
.